REGRID_LAS

Resamples a LiDAR point cloud onto a regular grid by converting the dataset to a raster DEM and then reconstructing a point cloud from that surface.

The workflow computes point density, derives a coverage mask, generates a DEM at the specified resolution, and converts the DEM back into a regularly spaced point cloud. The output represents a gridded surface with uniform point spacing, which is easier to visualize and process than irregular LiDAR data.

Typical use: converting bathymetric (MBES) or terrain LiDAR into a uniform surface representation before visualization or tiling. Especially useful when point density is highly variable across the dataset.

---

Contract

Type	REGRID_LAS
Accepts	input_las: las
Produces	output_las: las
Params	resolution (number)

---

Inputs

Slot	Type	Description
input_las	las	Source point cloud dataset

Outputs

Slot	Type	Description
output_las	las	Regularly spaced point cloud derived from the reconstructed surface

Parameters

Name	Type	Description
resolution	number	Grid cell size used when generating the DEM, in the dataset's spatial units

---

What it does internally

1. Compute point density raster

A PDAL pipeline reads the input and produces a density raster (density.tif) counting points per grid cell. Ground-classified points are excluded.

2. Extract coverage footprint

The density raster is processed through a sophisticated coverage detection pipeline:

1. Threshold pixels where density > 0
2. Polygonize the result (gdal_polygonize)
3. Union all polygons (ST_Union)
4. Remove small isolated areas below an area threshold
5. Smooth polygon edges with a double-buffer (ST_Buffer(+5) then ST_Buffer(-5))
6. Simplify geometry (ST_SimplifyPreserveTopology)

The result is a clean coverage mask (mask.gpkg) that defines the valid LiDAR footprint.

3. Generate DEM

A PDAL pipeline converts the input point cloud to a raster DEM (dem.tif) at the specified resolution, using mean elevation per cell.

4. Convert DEM back to point cloud

A PDAL pipeline reads the DEM raster and writes it as a LAS point cloud, producing a regularly spaced grid of points at the DEM's cell resolution.

The output point cloud represents the dataset's surface at uniform spacing — irregular point density from the original survey is removed.

---

Recipe usage

{
  "id": "regrid",
  "type": "REGRID_LAS",
  "inputs":  { "input_las": "job:input_las" },
  "outputs": { "output_las": "step:regrid.output_las" },
  "param_keys": ["resolution"]
}

Job params:

"params": {
  "regrid": {
    "resolution": 0.5
  }
}

---

Artifact storage path

artifacts/job_{id}/regrid/output.las

---

Notes

The resolution parameter is in the spatial units of the input dataset's CRS. If the dataset is in feet, resolution: 1 means 1-foot cells. Reproject first if a metric resolution is needed.